Autonomous Reality Modelling for Cultural Heritage Sites employing cooperative quadrupedal robots and unmanned aerial vehicles

Problem statement During recent years, Reality Modeling (RM) technologies including cutting-edge sensor technologies like Terrestrial and Aerial (using drones) Laser Scanning, have found a wide and prominent purpose in the field of Cultural Heritage (CH) modeling, recording and management. However, RM of CH is a constant challenge for surveyors, since it is a manual-driven, laborious and time-consuming process. The scanning path and sensor's positioning are mostly depended on the surveyor's experience, intuition and perception, since an automatic and systematic procedure does not exist. Taking into consideration the natural environment that surrounds CH sites, the challenges become even more complex. Specifically, for the acquisition of a complete 3D Reality model of a large-scale cultural space, multiple manual terrestrial laser scans (TLS) and aerial scans with UAVs (drones) must be performed. In this manual procedure, the scanning path/strategy and the identification of the scanner position, or Next Best View task as presented in the literature, is mostly depended on the operator's experience and perception. As a result, an optimization of the NBV problem to capture efficiently a large-scale complex sites or monuments in dynamic environments (e.g., due to growing or changing vegetation) is quite important to minimize the surveying time and scanning cost. Although the NBV problem is crucial, efficiency and optimality have not been considered qualitatively and explicitly so far in the literature, and thus, in some cases, the surveying process takes usually longer than necessary, since some regions are overlapped unnecessary and extra positionings are planned just to be on the safe side.